1

Effect of Growth Techniques on the Structural, Optical and

Electrical Properties of Indium doped TiO2 Thin Films

Faisal S. Al mashary a, b, *, Suelen de Castro c, Arlon Fernando da Silva d, Jorlandio Francisco Felix d, Marcelo

Rizzo Piton c, Helder Vinícius Avanço Galeti c, Ariano De Giovanni Rodrigues c, Yara Galvão Gobato c, Noor Al

Saqri a, Mohamed Henini a, e, Maryam M. Al huwayz a, Abdulrahman M. Albadri f, Ahmed Y. Alyamani f, Hamad

A. Albrathen f, g, Sami A. Alhusaini f, Khalid M. Aljaber f, Ali Z. Alanazi f, Fahad S. Alghamdi f

a School of Physics and Astronomy, University of Nottingham, Nottingham, NG7 2RD, United Kingdom b Department of Physics, College of Science, Qassim University, Buraydah, 14452, Saudi Arabia c Departamento de Física, Universidade Federal de São Carlos, São Carlos 13560-905, Brazil

d Institute of Physics, Universidade de Brasília, Brasília, DF 70910-900, Brazil e UNESCO-UNISA Africa Chair in Nanoscience & Nanotechnology Laboratories. College of Graduate

Studies, University of South Africa (UNISA), Muckleneuk Ridge, PO Box 392, Pretoria, South Africa f National Center for Nanotechnology and Advanced Materials, King Abdulaziz City for Science and Technology,

Riyadh, 11442, Saudi Arabia g Physics and Astronomy Department, ALASeR/KAIN, King Saud University, Riyadh 11451, Saudi Arabia

* Corresponding author. School of Physics and Astronomy, University of Nottingham, Nottingham, NG7 2RD,

United Kingdom

E-mail address: faisal.almashary@nottingham.ac.uk (F. S. Al mashary)

Keywords: In-doped TiO2, Pulsed laser deposition, Sputtering, XRD, Photoluminescence,

Deep Level Transient Spectroscopy

Abstract

We have investigated the effect of the growth techniques on the structural, the

electrically and optically active defects in Indium doped TiO2 thin films grown by pulsed laser

deposition (PLD) and sputtering techniques. X-ray diffraction (XRD) and Raman spectroscopy

patterns revealed both rutile and anatase phases for the sputtering samples. On the other hand,

only the anatase phase was observed for the PLD samples. The photoluminescence (PL) spectra

have unveiled several peaks which were explained by defect related optical transitions.

Particularly, the PL bands are fully consistent with anatase/rutile TiO2 phases and the formation

of In2O3 during the preparation of our samples. It was also observed that at -4 V reverse bias,

the PLD samples have lower leakage currents (~1.4 x 10-7 A) as compared to the sputtering

samples (~5.9 x 10-7 A). In addition, the PLD samples exhibited lower ideality factors and

higher barrier heights as compared to those grown by sputtering. Finally, the Deep Level

CORE Metadata, citation and similar papers at core.ac.uk

Provided by Nottingham ePrints

2

Transient Spectroscopy (DLTS) measurements have shown only one defect in the PLD samples

whereas five defects have been detected in the sputtering samples. Therefore, our results

provide strong evidence that the PLD technique is better suited for the growth of In-doped TiO2

thin films.

1. Introduction

Transparent conducting oxides (TCOs) materials have wide applications in terms of

absorbing, transparency of the visible light and electrical conductivity. They are used in

optoelectronic, photovoltaic, photocatalytic and photoelectrochemical water splitting

applications [1]. TiO2 is a TCO material which has been extensively studied for photocatalysis

and photovoltaic applications. This wide interest in TiO2 is due to its efficient photoactivity,

high stability, low cost and non-toxicity [2-4]. The large band gap (3.0-3.4 eV) [5, 6] of TiO2

makes it only sensitive to UV light and therefore it is not efficient for solar cell applications.

Among a lot of approaches that have been used to modify the electrical and optical properties

of TiO2 in order to make it sensitive to visible light, metal ion doping has been proved to be

one of the best methods [7]. Different metals have been used as dopants in TiO2 to change its

electronic and optoelectronic properties [8]. Using indium as a dopant has attracted a lot of

attention as it can modify some properties of TiO2, such as band gap, surface composition and

charge transport.

Oxygen vacancies are well-known defects in TiO2 which can alter the geometric

structure and the chemical properties of the system [9]. In addition to the defects that may be

introduced into the forbidden band gap of TiO2 due to incorporating indium, the growth

methods could also provide defect levels within the band gap. Understanding the deep and

shallow level defects is essential for future devices [10]. Deep level transient spectroscopy

(DLTS) and photoluminescence (PL) are powerful tools to investigate these defects in

3

semiconductors. Particularly, by using DLTS, defects parameters such as the activation energy

of a defect, its capture cross section and its concentration can be determined. In this study, we

investigated the effect of growth techniques, namely pulsed laser deposition (PLD) and

sputtering, on structural, optical and electrical properties of indium doped TiO2 thin films

deposited on silicon substrates.

2. Sample Details

Two series of samples were prepared for this study using two different growth methods.

For the first series, a layer of 380 nm TiO2 thin film, measured using cross-section by scanning

electron microscopy (SEM, HITACHI S4500), was deposited on (100) n-type silicon substrate

using pulsed laser deposition (PLD/MBE 2100) from PVD products. The laser source was a

KrF excimer Laser (λ = 248 nm, pulse width 20 ns, and repetition rate = 10 Hz) operating at

350 mJ to ablate the target with two sets of pressure (under vacuum: 10-6 Torr and with oxygen

background pressure: 5×10−3 Torr). The growth temperature was fixed at 500 ˚C and the

distance from substrate to target was set at 55 mm. All PLD parameters which contribute to the

growth rate were kept constant during the operating time. This series of samples is named PLD

sample. In the second series, RF magnetron sputtering technique was used to deposit a 300 nm

thin film of TiO2 on (100) n-type silicon substrate. This series is labelled as sputtering sample.

For the RF sputtering system, the samples were deposited using a 2-inche TiO2 target at a

temperature of 500 ˚C in pure Argon ambient without addition of oxygen. The chamber was

evacuated to a high vacuum of less than 1×10-8 Torr. The substrate was rotated during the

deposition at a low speed to enhance the thickness uniformity of the films. Substrates were

mounted on 300 mm stainless steel rotating disk where the distance between target and

substrate was 150 mm. High purity argon gas (99.999%) was introduced at a rate of 22 sccm

as an inert gas for plasma. Before deposition, the samples were pre-sputtered in argon plasma

4

for 10 mins to remove any contaminants. The working pressure was set to 10-3 Torr and RF

power was fixed at 150 W. By thermal evaporation, layers of indium (50 nm thick) were

deposited on top of each sample set in order to keep the same doping concentration of indium

and investigate only the effect of the growth techniques on the electrical and optical properties

of In doped TiO2 thin films. In order to incorporate indium into the TiO2 lattice, both series of

samples were annealed under oxygen flow with a temperature ramp rate of 15°C/minute and a

dwelling temperature of 500 °C for 30 minutes. This annealing process has been found as the

best among many annealing processes and procedures.

A third series of control samples, which do not contain In, were grown by sputtering

and PLD techniques using the same procedures as described above. This will allow us to

investigate the effect of In doping. These samples are labelled here as Reference samples

To prepare the samples for electrical characterization, the devices were processed in the

form of circular mesas with a diameter of 900 μm. The back sides of the silicon substrates were

exposed to evaporated Aluminium to create back ohmic contacts. Ti/Au has been deposited by

thermal evaporation on top of the In-doped TiO2 layer to create Schottky contacts. Fig. 1 shows

the growth and fabrication steps.

Fig. 1 Growth and fabrication steps.

5

The photoluminescence (PL) spectra of In doped TiO2 thin films were investigated as

a function of temperature using a laser excitation wavelength of 325 nm. X-ray diffraction

(XRD) patterns were measured in a Bruker D8 Discover Diffractometer, using Cu Kα radiation

(λ = 1.5418 Å). The diffraction pattern was obtained at diffraction angles between 10º and 80º

with geometry 2θ (the glancing angle X-ray diffraction-GAXRD method) and the conventional

θ-2θ at room temperature. The Raman measurements were performed using a Renishaw micro

Raman inVia spectrometer under ambient conditions. A laser with wavelength of 514 nm was

used as an excitation source.

Current-voltage (I-V) measurements of both series of samples have been performed

using a Keithley 236 source measure unit. Capacitance-voltage (C-V) measurements were

investigated at 300 K with 1 MHz frequency using Boonton 7200 capacitance meter. DLTS

measurements were carried out using a He closed-loop cycle cryostat (Janis CCS-450) and

Boonton 7200 capacitance meter.

3. Results and Discussion

3.1 Structural and optical investigation

Fig. 2 shows the XRD pattern of the PLD and sputtering samples. The XRD θ-2θ pattern

of the sputtering reference sample (Fig. S1 in the supplement) exhibited diffraction peaks at

25.4°and 25.5° corresponding to the (011) and (110) lattice planes revealing the anatase and

rutile phases, respectively. Fig. 2(a) shows the XRD pattern of the In doped TiO2 sputtering

sample (In doping obtained by depositing and annealing an indium layer of 50 nm thickness).

The result shows the presence of anatase and rutile phases of TiO2, however, we can also notice

6

an additional signature of indium oxide phase with the main peaks at 22.5º, 30.6º, 35.6º, 47.3º,

51.1º, and 60.9º corresponding to the (211), (222), (400), (431), (440), and (622) lattice planes

of cubic In2O3 (JCPDS File no:06-0416), respectively. Additionally, the precipitation and

formation of a second phase may contribute to the increasing of trap defects, as will be shown

later using the DLTS technique. TiO2 thin film grown by sputtering technique composed of

rutile and anatase phases have been also observed by Huang at al. [11].

Fig. 2 Room temperature XRD pattern of In doped TiO2 thin films prepared by (a) sputtering and (b)

PLD.

A single rutile phase characteristic of TiO2 thin films is obtained for PLD reference sample

(Fig. S2 in the supplement). Fig. 2(b) shows the XRD pattern of PLD samples. The peaks are

in good agreement with the standard XRD patterns of TiO2 found in the literature [11]. It can

be seen that the peak R(110) strongly dominates over the other peaks related to rutile TiO2

structure, indicating the preferential orientation. However, in the In-doped TiO2 PLD sample

the presence of both indium oxide phase and TiO2 pure- rutile phase are observed (see Fig.

2(b)). From the XRD results, it is possible to conclude that for the sputtering reference sample

both rutile and anatase phases are present, while only rutile phase was detected in PLD

reference sample. Moreover, the presence of a layer of indium (50 nm thickness) causes the

formation of In2O3 phases in both the sputtering and PLD samples. However, the

polycrystalline character of the In2O3 phase is more intense in the sputtering samples. This

7

latter behaviour will have a strong influence on the electrical properties of the devices, as will

be shown later.

In order to confirm the XRD results, Raman studies have been performed. Fig. 3(a)

shows the Raman spectra of the In-doped TiO2 samples obtained by sputtering, where both

rutile and anatase phases are observed. The sputtering sample shows an intense peak at Eg=142

cm-1 from anatase phase. The inset of Fig. 3(a) shows a weak peak at B1g=399 cm-1 and Eg=638

cm-1 that are characteristics of anatase phase of TiO2 [12]. A single band at Eg=446 cm-1 from

TiO2 rutile phase is also observed. Finally, a weak vibrational band has been observed for cubic

In2O3 at 197 cm-1 which is assigned to the O–In–O bending modes of cubic In2O3. The most

intense peaks at 520cm-1 are due to the Raman signal from Si(100) substrates. The results from

XRD also show similar behavior. The same Raman bands are also observed in the sputtering

reference sample (Fig. S3 in the supplement).

Fig. 3 Raman spectrum of In-doped TiO2 thin films prepared by (a) sputtering and (b) PLD.

Fig. 3(b) shows the Raman spectra of PLD samples. As can be seen in Fig. 3(b), only two peaks

at Eg=446 cm-1 and Eg=610 cm-1 are observed which are related to rutile phase of TiO2 [13].

These results are in good agreements with XRD results, where only the rutile phase was

8

observed. The Raman spectrum for PLD reference samples (Fig. S4 in the supplement) showed

the same characteristics.

3.2 Current-voltage measurements

I-V measurements as a function of temperature (20-440 K with 20 K intervals) were

performed in both samples in order to determine the diode parameters such as ideality factor

(n), barrier height (ϕb) and series resistance (Rs). Fig. 4 shows the room temperature semi-

logarithmic plot of I-V characteristics for both samples. It can be seen from the figure that the

reverse current of the sputtering sample is higher than that of PLD sample. In particular, the

reverse bias leakage current at – 4 V for PLD and sputtering samples are 1.3 x 10-7 A and 5.9

x 10-7 A, respectively. This may indicate the presence of more defects in sputtering samples

than PLD samples. These defects which act as generation-recombination centres and contribute

to the tunnelling process in reverse bias can be probed by DLTS technique [8]. In addition,

from the linear I-V plot (inset of Fig. 4) a turn-on voltage (Von) of 0.26 V and 0.77 V was

obtained for the sputtering and PLD sample, respectively.

The I-V characteristics of an ideal diode including series resistance (Rs) could be

described by the thermionic emission model [14] as:

𝐼 = 𝐼0[𝑒𝑥𝑝 (𝑞(𝑉−𝐼𝑅𝑠)

𝑛𝐾𝑇) − 1] (1)

where 𝐼0 is the saturation current and is given by:

𝐼0 = 𝐴𝐴∗𝑇2 𝑒𝑥𝑝 (−𝑞𝜙𝑏

𝐾𝑇) (2)

In Eqs. (1) and (2) A is the effective diode area (𝐴 = 6.362 x 10-3 cm2 for both samples), 𝐴∗ is

the effective Richardson’s constant (𝐴∗ = 671 A cm-2 K-2 [15]), k is Boltzmann’s constant, T

is the temperature, q is the elementary charge, n is the ideality factor, ϕb is the barrier height

and Rs is the series resistance. In order to improve the accuracy of calculating the diode

9

parameters (n, ϕb, Rs and 𝐼0), the Werner’s method was used [16]. These parameters for PLD

and sputtering samples at room temperature are shown in Table. 1.

Fig. 4 Semi-logarithmic I-V plots for PLD and sputtering samples. The inset shows the linear I-V plot.

Table. 1 Ideality factor (𝑛), barrier height (ϕb) and series resistance (Rs) at room temperature for

PLD and sputtering samples.

Sample 𝑛 ϕ𝐵 (eV) R𝑆 (kΩ)

PLD 2.01 0.87 330.86

sputtering 2.54 0.77 5.92

The plot of ln (I0/T2) versus 1000/T for PLD sample (Fig. S5 (a) in the supplement) in

the temperature range 100-420 K follows a straight line. This behaviour suggests that the

conduction mechanism could be governed by thermionic emission and the saturation current

can be obtained using Eq. (2) [17]. A similar behaviour was observed in the sputtering sample

in the temperature range 160-420 K (Fig. S5 (b) in the supplement). The Richardson constants

that are calculated from the intercept of the straight line of ln (I0/T2) versus 1000/T plot for

10

PLD sample (A∗= 1.917 x 10-6A/K2 cm2) and sputtering sample (A∗= 4.956 x 10-8 A/K2 cm2)

are much lower than the well-established value (A∗= 671 A/K2 cm2). This deviation from the

Richardson constant value may be due to the spatial inhomogeneous barrier and potential

fluctuations at the interface that consist of low and high barrier areas [18, 19]. Fig. 5 shows

the temperature dependence of the barrier height (ϕ𝐵

) and ideality factor (𝑛) for both samples.

As the temperature increases, the barrier height increases while the ideality factor decreases.

This behaviour is known to be due to nonuniformity of interfacial charges. At low temperature,

carriers are frozen and therefore the current does not follow the thermionic emission

mechanism. At low temperatures, the current passes through the interface states and this result

in higher values of the ideality factor [20]. Note that the carriers, at low temperature, can

overcome the lower barriers and the mechanism of transport will be dominated by the current

flowing through the region with lower barrier height. As temperature increases, carriers gain

sufficient energy to overcome the higher barriers, which results in an increase of the barrier

height with temperature [21].

The barrier height and ideality factor behaviour with temperature have been attributed

to the barrier inhomogeneity. This inhomogeneity of barrier height was also confirmed by the

almost linear relationship between barrier height and ideality factor for both samples (Fig. S6

in the supplement) [22]. It is worth noting that deviation from a linear relationship has been

attributed to uncertainties in the calculation of the ideality factor [23].

11

Fig. 5 In doped TiO2 samples grown by PLD and sputtering (a) barrier height (ϕ𝐵

) versus temperature;

(b) ideality factor (𝑛) versus temperature.

The main trap which contributes to the leakage current can be calculated using I-V

measurements at different temperatures by plotting the reverse current versus inverse

temperature [20]. Fig. 6 shows the plot of reverse current versus 1000/T for -1 V bias voltage

for both samples. One main trap has been found in PLD sample with an activation energy

EPLD= 0.492 ± 0.007 eV. However, two traps have been detected in sputtering sample, namely

ESP1= 0.100 ± 0.005 eV and ESP2= 0.51 ± 0.04 eV. Note that EPLD and ESP2 have similar

activation energies and the origin of this trap may be due to oxygen vacancy [7] or a deep donor

Ti+3 state [10]. ESP1 may be related to oxygen vacancies or interstitial Ti ion [24]. A large

number of defects determined from I-V measurements could be explained a high value of

reverse current [20]. Since the sputtering sample has a larger number of defects than the PLD

sample that could account of the larger value of the reverse current observed in the sputtering

sample. It is worth noting that the numbers of traps found using the I-V measurements are

consistent with those found by using optical technique, as we will show later.

12

2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5

-20

-18

-16

-14

-12

-10

VR= -1V

EP= (0.492 ± 0.007) eV

ES1= (0.1 ± 0.005) eV

ES2= (0.51 ± 0.04) eV

Linear Fit

Linear Fit

Linear Fit

ln(I

)

1000/T (K-1)

Fig. 6 Reverse current versus 1000/T for reverse bias voltage VR = -1 V for In-doped TiO2 samples

prepared by PLD and sputtering

3.3 Capacitance-voltage measurements

The C-V measurements for both samples at room temperature are shown in Fig. 7(a).

It is clear from the figure that the sputtering sample has higher capacitance than the PLD sample

and this could be due to shallow ionised impurity levels. Also, a small shoulder can be seen in

both samples, at -0.6 V and at -1.3 V for PLD and sputtering, respectively. This behaviour

could be due to a poor interface quality and drift in carriers [8]. The slope of the best fit of the

plot 1/C2 versus reverse bias, Fig. 7(b), was used to calculate the free carrier concentrations,

Nd, for both samples at room temperature. For the PLD sample, the doping concentration was

found to be uniform as evidenced by the linear behaviour of the plot of 1/C2 versus applied

bias. The free carrier concentration is about 5.4 x 1013 cm-3. However, as depicted in Figure 7

(b), for the sputtering sample, two lines were fitted to the C-V data, confirming two different

13

doping concentrations of 1.0x1014 cm-3 and ~3.0x1014 cm-3. The free carrier concentration is

higher near the junction interface than in the bulk.

Fig. 7 (a) C-V plot and (b) 1/C2 versus V plot at room temperature of In-doped TiO2 samples prepared

by PLD and sputtering.

14

3.4 Optical properties

Fig. 8 (a) and (b) show typical normalized PL spectra at 12K and 300K for both samples

prepared by sputtering and PLD. It was observed that the PL spectrum of the sputtering sample

consists of only one emission broad band around 1.95 eV. This result is similar to the one

previously reported for In doped TiO2 films prepared by e-beam evaporation [8]. The nature

of this emission band was previously attributed to the In3+ cation, which may produce a

radiative transition between In (5P) and oxygen defect level [8]. However, the observed broad

PL band could also be associated to optical transitions related to anatase and rutile phases. It

can also include an important contribution of In2O3 which can be formed during the thermal

annealing of the samples to incorporate In. On the other hand, the PLD sample has revealed

several peaks. In order to understand the contribution of In doping we have also studied PL

from reference samples prepared by sputtering and PLD. Fig. 8 (c) and (d) show the normalized

PL spectra at 300K for the reference and doped TiO2 samples prepared by both techniques. We

remark that for the reference sample prepared by sputtering the PL spectra shows four emission

peaks (Fig. 8(d)) at 300K. As will be discussed below, the anatase phase of TiO2 usually shows

PL bands around 1.9 and 2.4eV while the rutile phase usually shows PL bands around 1.5 and

1.7eV [25]. Therefore, the PL peaks observed at 300K for the reference sample prepared by

sputtering are fully consistent with the presence of anatase and rutile phases which were also

evidenced by XRD and Raman results.

As mentioned previously several bands were observed for the PLD sample in the visible

and infrared region at 12K and 300K (Fig. 8(a) and 8(b)) which are labelled as P1 (1.93 eV),

P2 (2.52 eV), P3 (2.90 eV), and P4 (3.25 eV) at 12 K. In the following, we present a detailed

discussion about the nature of these emission bands and possible interpretation of PL peaks

observed in our PLD sample.

15

Fig. 8 Normalized PL spectra of In-doped TiO2 thin films grown by PLD and sputtering at (a) 12K

and (b) 300K. Comparison between PL emission at 300K for undoped and In-doped TiO2 grown by (c)

PLD and (d) sputtering.

As mentioned previously the PL spectra of TiO2 material usually depends on the crystal

phase (anatase, rutile or mixed phases) [25]. It was reported [25] that the TiO2 PL spectrum

shows green (around 2.5 eV) and red (around 1.9 eV) emissions for the anatase phase and only

a near-infrared emission (NIR) band around 1.5 eV for the rutile phase. The green PL emission

is usually attributed to self-trapped excitons [26], oxygen vacancies [27] and surface states

[28]. However, a detailed study of these emissions has demonstrated that the green emission

could also be related to radiative recombination of free electrons with trapped holes. These

trapped holes could be localized at oxygen vacancies sites (VO) or at Ti sites adjacent to VO

[25]. The red PL emission is usually related to recombination between trapped electrons and

free holes [25]. On the other hand, the nature of NIR emission in TiO2 is more difficult to

explain [25]. Usually, the rutile samples are less sensitive to prolonged exposure to UV

16

illumination in oxygen environment than anatase samples, which suggests that the NIR PL

could be associated to a possible radiative recombination between midgap trapped electrons

and free holes at valence band. On the other hand, our XRD and Raman spectroscopy results

have revealed both TiO2 and In2O3 phases for the In-doped TiO2 samples. The formation of

In2O3 after the thermal annealing process which was used to incorporate In in the TiO2 lattice

could also contribute to the PL bands observed in the In-doped TiO2 samples. It is well known

that In2O3 oxide is also a wide band gap (~3.6 eV) semiconductor [29]. It was previously

reported that the PL spectra displayed a near band edge (NBE) emission around 3.23 eV and a

defect-related emission at ~1.98 eV [30]. The PL peak in the visible region ( defect-related

emission) was previously associated to the transition from the oxygen vacancy level to the

valence band[31]. In addition, this red emission was also previously associated with deep level

states (1.94 eV) which can be present in In2O3 [32].

As mentioned above our In-doped TiO2 PLD samples revealed four emission bands

(labelled P1, P2, P3 and P4). It was found that as the temperature increases, a gradual reduction

of P1 emission band was observed as expected. However, additional weak emissions bands

located around 1.5eV and 1.6eV were also detected. The PL peak energies of most of these

bands are smaller than the energy gap of TiO2 and In2O3. Therefore, all these emission bands

were attributed to the localized levels in the bandgap and could be due to intrinsic defects in

both TiO2 and In2O3. As mentioned above, the observed defect emission peaks for TiO2 and

In2O3 usually have emissions in the same energy range. Therefore, it is difficult to separate the

contribution of defect-related emissions from the In2O3 and TiO2. In order to investigate which

peak could be associated to the TiO2, we have also studied a reference sample, i.e. undoped

TiO2 sample. Fig. 8 (c) shows the 300 K PL spectra of undoped and In-doped TiO2 50nm

samples prepared by PLD.

17

As can be seen in Fig. 8(c), the PL peaks of the undoped TiO2 thin film have a smaller

intensity in the range of about 1.8-2.9 eV and around 1.35 eV as compared to those in In-doped

TiO2 samples. The observed bands of 1.8-2.9 eV and ~1.35 eV are assigned to the anatase and

rutile phases (mixed phase), respectively [25]. The visible emission was associated to a

combination of two bands previously reported in the literature for the anatase phase, namely

the green band (~2.5 eV) and the red band (~1.9 eV) [25]. However, the anatase phase was not

evidence for XRD and Raman results. This band could be associated to small contribution

anatase phase (mixed phase sample) that cannot be evidenced by XRD and Raman results but

it is important for optical emission. On the other hand, emissions around 1.5 and 1.7 eV have

been observed only for the rutile phase and associated with the radiative recombination

between midgap trapped electrons and free holes at the valence band [25] and Ti3+ ions [33].

We attribute the broad band around 1.2-1.6eV to the rutile phase. On the other hand, additional

PL peaks (P1, P2, P3 and P4) were observed in PLD sample which were associated to the

formation of In2O3. Particularly, the P4 band that observed around 3.25 eV at 12 K was

associated to the band edge emission of In2O3 [32]. The other PL peaks (P1, P2 and P3) have

important contribution of defect emission due to In2O3.

18

We have also investigated the temperature dependence of the emission bands for both

PLD and sputtering samples. Figs. 9 (a) and (b) show a typical temperature dependence of PL

spectra for these samples. It was found that the emission intensity of the peak at 1.93 eV (P1)

decreases considerably with the increase of temperature for both sputtering and PLD samples

(please see Fig. 9(a)) as expected. However, the intensities of the other emissions of the PLD

sample increase when the temperature increases (in the range of 100 to 300K). We believe that

this effect could be associated to charge transfer, or traps due to the presence of defects or some

disorder in TiO2 and/or In2O3. Traps could localize carriers non-radiatively, and the trapped

carriers are thermally activated and contribute to the optically active states. We have also

estimated the activation energy using a double-channel activation energies function [34, 35].

𝐼(𝑇) =𝐼0

1+𝛾1exp (−𝐸𝑎1/(𝑘𝐵𝑇))+𝛾2exp (−𝐸𝑎2/(𝑘𝐵𝑇)) (3)

where 𝐼𝑜 is the PL intensity at T = 0 K, 𝛾1,2are two constants related to the ratio between

radiative and non-radiative recombination processes, 𝐸𝑎1,2 are the activation energies

corresponding to the non-radiative recombination process, and 𝑘𝐵 is the Boltzmann constant.

19

The inset in Fig. 9(a) and (b) shows the temperature (1/kBT) dependence of the integrated PL

intensity (I) (Arrhenius plot) in the temperature range T = 12–300 K for PLD sample (P1=1.93

eV) and sputtering sample (peak at 1.95eV). Clearly, the emission intensity decreases with

increasing temperature following an Arrhenius-like function. Equation (3) has been used to fit

the PL intensity in the whole measured temperature range (dash line). Activation energies of

𝐸𝑆𝑃𝑎1 =82.0 meV and 𝐸𝑆𝑃𝑎2 =12.0 meV, and 𝐸𝑃𝐿𝐷𝑎1 =22.3 meV were obtained for sputtering

and PLD thin films, respectively. Al Saqri et al. [8] reported a defect in In-doped TiO2 thin

films with an activation energy of 62 meV. They attributed this defect to the intrinsic Ti atoms

and ionization energies of oxygen vacancies. In our case, the calculated activation energy could

also be related to the similar nature which has already been evidenced from emission processes.

Fig. 9 PL spectra of (a) sputtering sample and (b) PLD sample at different temperatures in the range

12-300K. The insert shows the Arrhenius plot of total intensity for both samples.

20

In conclusion, we have observed well resolved PL bands in the PLD sample and a broad

PL band in the sputtering sample. These results provide strong evidence that PLD samples are

of higher quality than sputtering samples. However, important contribution defect emission

from the formation of In2O3 was observed in the PL from PLD samples. These results are also

consistent with the I-V results for PLD samples which have shown better performance than the

sputtering samples. The optical results are also consistent with the DLTS results which will be

presented below.

3.5 DLTS and Laplace DLTS measurements

In order to investigate the effect of the growth techniques on the electrically active

defects, DLTS technique has been used [36]. The experiment was carried out with a reverse

bias, VR = -1V, filling pulse height, VP = 0V, filling pulse time, tP = 1 msec and rate windows

200 s-1 for both samples. Fig. 10 shows the DLTS signals versus temperature for both samples

over the scanned temperature range 10K-450 K. Broad peaks due to electron traps have been

detected in both samples as shown in Fig. 10. Laplace DLTS measurements [37] have been

carried out on both samples in order to resolve the broader peaks. In the Laplace DLTS spectra

(Fig. S7 in the supplement) only one electron trap (EP) and five electron traps (Es1, Es2, Es3, Es4,

and Es5) were detected in the PLD and sputtering samples, respectively. The activation energies

of these traps were calculated from the Arrhenius plots as shown in Fig. 11. The trap

parameters, such as activation energies, trap concentrations and capture cross sections are

summarised in Table.2.

As seen in Table. 2, traps Ep and Es4 have the same energies so they may have the same

origin which could be assigned to oxygen vacancy as reported by E. Wang et al. [7] in In-doped

TiO2 doped using the sol-gel method. However, B. Morgan and G. Watson have investigated

the formation of native defects in anatase TiO2 using the density functional theory [10] and

21

assigned a trap at 0.5 eV to Ti+3 state. Ep and Es4 could have the same origin. Es1 is a shallow

level trap and Es2 may be related to oxygen vacancy or interstitial Ti ion [24]. The Es2 trap was

also observed using PL measurement, labeled as 𝐸𝑆𝑃𝑎1. Es3 and Es5 may be assigned to a Ti

donor level [38].

Fig. 10 DLTS signal for PLD and sputtering samples.

Fig. 11 Arrhenius plots from Laplace DLTS with reverse biases, VR= -1 V, filling pulse height, VP =

0V, and filling pulse time, tP = 1 msec, for (a) PLD sample and (b) sputtering sample

22

Table. 2 Traps parameters for PLD and sputtering samples at VR= -1V, VP= 0V and tP= 1 msec.

DLTS measurements using different reverse biases are usually performed to control the

depletion region width (moving it away or close to the interface) where the process of trapping

and de-trapping of carriers happen. DLTS measurements have been carried out for PLD sample

at different reverse biases VR = -1, -3, and -4V, as shown in Fig. 12. The filling pulse height,

VP = 0V and the filling pulse time, tP = 1 msec, were kept the same for all measurements. As

we change the reverse bias VR, only one trap has been detected for each reverse bias. Table 3

shows the trap parameters for VR= -1, -3 and -4 V. Arrhenius plots for VR= -3 and -4 V are

displayed in Fig. 13. The trap found at VR= -3 V has the same activation energy as that at VR=

-4 V whose origin could be related to the Ti donor [38]. Note that the amplitude of the DLTS

signal is decreasing as the reverse bias increases (away from the interface). This means that the

concentration of the trap is decreasing as moving away from the interface.

Sample

Trap

Activation

Energy

(eV)

Capture Cross-

Section (σ∞)

(cm2)

Trap Concentration

(cm−3)

PLD Ep 0.469 ± 0.009 3.998 x 10-16 3.749 x 1013

sputtering

Es1 0.0027 ± 0.0001 1.958 x 10-22 5.027 x 1012

Es2 0.096 ± 0.005 6.629 x 10-21 3.939 x 1013

Es3 0.21 ± 0.02 4.013 x 10-17 9.187 x 1013

Es4 0.415 ± 0.001 5.919 x 10-13 1.585 x 1014

Es5 0.324 ± 0.003 8.975 x 10-18 1.341 x 1014

23

Table. 3 Traps parameters for PLD sample at VR= (-1, -3 and -4) V, VP= 0V and tP= 1 msec.

Reverse bias

(V)

Trap

Activation Energy

(eV)

Capture Cross-

Section (σ∞)

(cm2)

Trap

Concentration

(cm−3)

-1 Ep 0.469 ± 0.009 3.998 x 10-16 3.749 x 1013

-3 Ep1 0.335 ± 0.003 1.171 x 10-17 2.854 x 1013

-4 Ep1 0.318 ± 0.002 7.779 x 10-18 2.101 x 1013

Fig. 12 DLTS signal for PLD sample at VR = -1, -3, and -4V

24

Fig. 13 Arrhenius plots for PLD sample obtained from Laplace DLTS at (a) VR= -3 V and (b) VR= -4

V.

4. Conclusion

The effect of the growth technique on the structural, electrically and optically active

defects in In-doped TiO2 has been investigated using two TiO2 sample sets prepared by PLD

and sputtering techniques. The samples grown by sputtering technique have shown more

defects than those grown by PLD. This could be related to the coexistence of both anatase and

rutile phases present in the thin film deposited by sputtering. As a consequence, the I-V

characteristics show lower reverse current for PLD sample as compared with the sputtering

sample. PLD sample has a better value of ideality factor than sputtering sample. One defect

has been found in PLD sample set where five defects have been detected in the sputtering

sample set. Due to the lower leakage currents and less number of defects in PLD sample, it is

concluded that the PLD technique is better suited for the growth of In-doped TiO2.

Acknowledgements

The author would like to thank Qassim University, Saudi Arabia, for funding his PhD

studies. The authors also acknowledge the financial supports from the Brazilian agencies

(Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP grants 2016/10668-7 and

25

2014/50513-7) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

(grant 3181/2014-1) and (FAPDF)17077.78.35088.2504/2017.

26

References

1. Stadler, A., Transparent conducting oxides—An up-to-date overview. Materials,

2012. 5(4): p. 661-683.

2. Hashimoto, K., H. Irie, and A. Fujishima, TiO2 photocatalysis: a historical overview

and future prospects. Japanese journal of applied physics, 2005. 44(12R): p. 8269.

3. Sarkar, M.B., et al., Enlarged broad band photodetection using Indium doped TiO2

alloy thin film. Journal of Alloys and Compounds, 2014. 615: p. 440-445.

4. Chinnamuthu, P., et al., Band gap enhancement of glancing angle deposited TiO2

nanowire array. Journal of Applied Physics, 2012. 112(5): p. 054315.

5. Amtout, A. and R. Leonelli, Optical properties of rutile near its fundamental band

gap. Physical Review B, 1995. 51(11): p. 6842.

6. Tang, H., et al., Urbach tail of anatase TiO 2. Physical Review B, 1995. 52(11): p.

7771.

7. Wang, E., W. Yang, and Y. Cao, Unique surface chemical species on indium doped

TiO2 and their effect on the visible light photocatalytic activity. The Journal of

Physical Chemistry C, 2009. 113(49): p. 20912-20917.

8. Al Saqri, N.A., et al., Investigation of defects in indium doped TiO 2 thin films using

electrical and optical techniques. Journal of Alloys and Compounds, 2017. 698: p.

883-891.

9. Nowotny, J., T. Bak, and M.A. Alim, Dual mechanism of indium incorporation into

TiO2 (Rutile). The Journal of Physical Chemistry C, 2014. 119(2): p. 1146-1154.

10. Morgan, B.J. and G.W. Watson, Polaronic trapping of electrons and holes by native

defects in anatase TiO 2. Physical Review B, 2009. 80(23): p. 233102.

11. Huang, H.-H., et al., Preparation of rutile and anatase phases titanium oxide film

by RF sputtering. Journal of nanoscience and nanotechnology, 2008. 8(5): p. 2659-

2664.

12. Mikami, M., et al., Lattice dynamics and dielectric properties of TiO 2 anatase: a

first-principles study. Physical Review B, 2002. 66(15): p. 155213.

13. Lukačević, I., et al., Lattice dynamics and Raman spectrum of rutile TiO2: The role

of soft phonon modes in pressure induced phase transition. Materials Chemistry

and Physics, 2012. 137(1): p. 282-289.

14. Sze, S.M. and M.K. Lee, Semiconductor devices : physics and technology. 2013,

Singapore: Wiley.

15. Pillai, P., et al., Extraction of Schottky barrier at the F-doped SnO2/TiO2 interface

in Dye Sensitized solar cells. Journal of Renewable and Sustainable Energy, 2014.

6(1): p. 013142.

16. Werner, J.H., Schottky barrier and pn-junctionI/V plots—Small signal evaluation.

Applied Physics A: Materials Science & Processing, 1988. 47(3): p. 291-300.

17. Van, C.N. and K. Potje-Kamloth, Electrical and NO x gas sensing properties of

metallophthalocyanine-doped polypyrrole/silicon heterojunctions. Thin Solid Films,

2001. 392(1): p. 113-121.

18. Jameel, D., et al., Electrical performance of conducting polymer (SPAN) grown on

GaAs with different substrate orientations. Applied Surface Science, 2016. 387: p.

228-236.

19. Aydoğan, Ş., M. Sağlam, and A. Türüt, On the barrier inhomogeneities of

polyaniline/p-Si/Al structure at low temperature. Applied surface science, 2005.

250(1): p. 43-49.

20. Aziz, M., et al., Electrical Behavior of MBE Grown Interfacial Misfit GaSb/GaAs

Heterostructures With and Without Te-Doped Interfaces. IEEE Transactions on

Electron Devices, 2015. 62(12): p. 3980-3986.

21. Tripathi, S. and M. Sharma, Analysis of the forward and reverse bias IV and CV

characteristics on Al/PVA: n-PbSe polymer nanocomposites Schottky diode. Journal

of Applied Physics, 2012. 111(7): p. 074513.

27

22. Korucu, D., A. Turut, and H. Efeoglu, Temperature dependent I–V characteristics

of an Au/n-GaAs Schottky diode analyzed using Tung’s model. Physica B:

Condensed Matter, 2013. 414: p. 35-41.

23. Al Saqri, N., et al., Investigation of the effects of gamma radiation on the electrical

properties of dilute GaAs 1− x N x layers grown by Molecular Beam Epitaxy. Current

Applied Physics, 2015. 15(10): p. 1230-1237.

24. Miyagi, T., et al., Photocatalytic property and deep levels of Nb-doped anatase TiO2

film grown by metalorganic chemical vapor depostion. Japanese journal of applied

physics, 2004. 43(2R): p. 775.

25. Pallotti, D.K., et al., Photoluminescence Mechanisms in Anatase and Rutile TiO2.

The Journal of Physical Chemistry C, 2017. 121(16): p. 9011-9021.

26. Skowronski, L., R. Szczesny, and K. Zdunek, Optical and microstructural

characterization of amorphous-like Al 2 O 3, SnO 2 and TiO 2 thin layers deposited

using a pulse gas injection magnetron sputtering technique. Thin Solid Films, 2017.

632: p. 112-118.

27. Iijima, K., et al., Influence of oxygen vacancies on optical properties of anatase TiO

2 thin films. Journal of Luminescence, 2008. 128(5): p. 911-913.

28. Yang, Y., et al., Visible and near-infrared electroluminescence from TiO2/p+-Si

heterostructured device. AIP Advances, 2014. 4(4): p. 047109.

29. Kim, H.S., et al., High performance solution-processed indium oxide thin-film

transistors. Journal of the American Chemical Society, 2008. 130(38): p. 12580-

12581.

30. Chan, C.-H., et al., Optical characterization of structural quality in the formation of

In2O3 thin-film nanostructures. The Journal of Physical Chemistry C, 2016.

120(38): p. 21983-21989.

31. Meléndez, J.J. and M. Wierzbowska, In2O3 doped with hydrogen: electronic

structure and optical properties from the pseudopotential Self-Interaction

Corrected Density Functional Theory and the Random Phase Approximation. The

Journal of Physical Chemistry C, 2016. 120(7): p. 4007-4015.

32. Ghosh, A., et al., Detailed investigation of defect states in Erbium doped In2O3 thin

films. Materials Research Bulletin, 2018. 99: p. 211-218.

33. Johannsen, S.R., et al., Influence of TiO 2 host crystallinity on Er 3+ light emission.

Optical Materials Express, 2016. 6(5): p. 1664-1678.

34. Krustok, J., H. Collan, and K. Hjelt, Does the low-temperature Arrhenius plot of the

photoluminescence intensity in CdTe point towards an erroneous activation energy?

Journal of applied physics, 1997. 81(3): p. 1442-1445.

35. Wang, X., et al., Trap states and carrier dynamics of TiO 2 studied by

photoluminescence spectroscopy under weak excitation condition. Physical

Chemistry Chemical Physics, 2010. 12(26): p. 7083-7090.

36. Lang, D., Deep‐level transient spectroscopy: A new method to characterize traps in

semiconductors. Journal of applied physics, 1974. 45(7): p. 3023-3032.

37. Dobaczewski, L., A. Peaker, and K. Bonde Nielsen, Laplace-transform deep-level

spectroscopy: The technique and its applications to the study of point defects in

semiconductors. Journal of Applied Physics, 2004. 96(9): p. 4689-4728.

38. Salama, A. and L. Cheng, The Effects of Titanium Impurities in N+/P Silicon Solar

Cells. Journal of The Electrochemical Society, 1980. 127(5): p. 1164-1167.